Predict, observe, conclude
The three words look like a worksheet heading and are actually the method that separates science from opinion. Most marks lost on investigation questions come from blurring them together.
Why insist on saying what you expect first?
Because a prediction made afterwards is not a prediction, and a human being is remarkably good at not noticing they have made one.
The effect has a name — hindsight bias — and it is not a character flaw. Once you know how something turned out, the result feels like what you expected all along, and the memory genuinely rewrites itself. Writing the prediction down before looking is the only reliable defence, which is why every serious study registers its hypothesis in advance.
The second discipline is separating what you saw from what you think it means. “The ice got smaller” is an observation. “The ice melted because of the lamp” is a conclusion, and it may be wrong even though the observation is right — perhaps the room was warm. Keeping them in separate sentences is what makes it possible to find that out.
Alberta adds this in Grade 1 and keeps asking for it every year, because it is the part of science that is a habit rather than a fact.
Where it turns up
- Medicine — a drug trial registers its hypothesis in advance, specifically to prevent hindsight bias
- Reading a news story — “studies show” is worth nothing until you ask what was predicted and what was measured
- Debugging anything — guess, test, and be told you are wrong — the same loop, faster
- Any lab report you will ever write — the marks are distributed exactly along these three steps
- 1. The words, first
- 2. The three steps, kept apart
- 3. Variables, and what makes a test fair
- 4. Saying how sure you are
- 5. Writing it up
The words, first
The idea: Six terms, and the exam distinguishes between all of them.
| Word | What it means |
|---|---|
| Hypothesis | A testable, falsifiable explanation. If no possible result would count against it, it is not one. |
| Prediction | What you expect to observe, said before observing. Often phrased as if–then. |
| Observation | What was actually measured or seen. No interpretation, no causes. |
| Inference / conclusion | What the observations suggest. This is where explaining is allowed. |
| Independent variable | The one you deliberately change. |
| Dependent variable | The one you measure, which may respond. |
| Controlled variable | Everything deliberately kept the same, so the change you measure has one plausible cause. |
| Fair test | An investigation in which only the independent variable changes. |
The three steps, kept apart
The idea: Each one has its own job, and doing two at once is what costs marks.
Predict. “If the lamp is moved closer, then the ice will melt faster, because more energy arrives each second.” Note the because — a prediction with a reason is testable in a way that a bare guess is not.
Observe. “At 10 cm the ice was gone in 4 minutes; at 30 cm it took 11 minutes.” Numbers, units, no explanation.
Conclude. “Melting was faster when the lamp was closer, which supports the prediction. Energy arriving per second falls with distance.”
The word supports is doing real work. One experiment does not prove anything, and saying so is a mark rather than a hedge.
Variables, and what makes a test fair
The idea: Change one thing, measure one thing, hold everything else still. Naming which is which is often worth a mark on its own.
Worked example. Does the colour of a surface affect how much it warms in sunlight?
- Independent: the colour of the surface.
- Dependent: the temperature after a fixed time.
- Controlled: the material, the size, the distance from the lamp, the starting temperature, the time, the thermometer.
Why controls matter more than they look. If the dark sample is also thicker, a difference in temperature has two possible causes and the experiment cannot distinguish them. The result is not wrong; it is uninterpretable, which is worse.
Repeat trials. One measurement could be a fluke. Three that agree are evidence; three that disagree are also evidence — that something uncontrolled is varying.
Saying how sure you are
The idea: Every measurement has an uncertainty, and every conclusion has a limit. Stating both is a sign of understanding rather than of doubt.
Measurement uncertainty. A length recorded as 7.0 cm to the nearest tenth could truly be 6.95 to 7.05. Writing 7 cm claims much less precision, and dropping a trailing zero quietly throws information away.
What an experiment cannot tell you. A correlation between two things does not establish that one caused the other. A third factor may cause both, or the effect may run the other way. Naming that possibility when it exists is worth more than ignoring it.
Sources of error, usefully described. “Human error” earns nothing. “The stopwatch was started by eye, so each time could be out by a few tenths of a second, which matters most for the shortest trial” earns marks, because it names the mechanism and says what it affected.
Writing it up
The idea: The structure is the same every time, and the marks follow it.
- Question. What are you trying to find out?
- Hypothesis and prediction. With a reason, and written before.
- Variables. Independent, dependent, controlled — listed.
- Method. Detailed enough that someone else could repeat it and get the same result. That is the actual test of a method.
- Observations. A table, with units in the headings.
- Analysis. A graph where there is a relationship to see.
- Conclusion. Does it support the prediction? What are the limits?
The single most common lost mark is a conclusion that restates the observations without answering the question. Say what it means, and say what it does not show.